Open source, self-hosted, MIT

Your database is a universe.

Fly through every table and relationship as a living map, then query it with SQL that knows your foreign keys.

Get DBZuper

See the shape of your data before you read a single row.

Tables become stars, sized by row count. Foreign keys become light. Related tables gather into coloured clusters, so a 400-table schema reads like a map instead of a list. Hover or tap a star: each one is a table from a sample shop.

  • Declared and inferred relationships, from *_id naming
  • Layouts saved per connection; drag a table to pin it
  • Stays smooth on 2,000-table schemas

Zoom is semantic. The closer you get, the more it tells you.

Cluster names fade into table names. Hover one to light up its neighbours, double-click to focus on a neighbourhood, Ctrl K to jump anywhere.

Right down to the columns.

Up close, every table opens into a column card: keys, types, references. One more click and you're in its data, its structure, or a SQL tab with the query already written.

Try the path finder

Universe, data and SQL. One window.

DBZuper showing the Schema Universe graph, a filtered orders table and the SQL editor suggesting a foreign-key JOIN

Only in DBZuper

Pick two tables. Get the JOIN.

How does a customer connect to a warehouse? DBZuper walks the relationship graph, shows you the shortest chain, and writes the JOIN for you, inferred hops included.

Click a start table, then an end table.

shop · PostgreSQLFind path to…
Ready-to-copy JOIN 

          

        

      

    


    
    

      

        
Only in DBZuper

        
Follow one row wherever it leads.

        
Every record is a card with a port per relationship. Click a port and the related rows appear next to it.
          The whole trail lives in the URL, so you can paste it to a teammate, or turn it into SQL.

      

      

        

          
dbzuper.local/c/shop/records?r=

          

            
            
          

        

        

          

            
            

          

        

        
      

      
Try it: open orders, then an order's order_items, then the product. Dashed ports are inferred relationships.

    


    
    

      

        
And when you need rows, a real grid.

        
Filter, sort and page any table. It all runs in the database, not in your browser, and one click turns the view into SQL.

      

      

        

          
 orders table

          
          
        

        

          
          
          
          
          
          
          
          
        

        

        

          

            

            
          

        

        

          
          

            
            
            
          

        

        

      

      
Click a header to sort, add a filter, or click a customer_id to follow the foreign key. Sample data.

    


    
    

      

                
Autocomplete for JOINs. A safety net for DELETE.

        
Monaco with schema-aware autocompletion that suggests whole foreign-key JOINs. And a safety net
          that makes you mean it before anything destructive runs.

      


      

        

          

            
 RunRun allExplain

            orders.sql
          

          

            
            

            
          

          

            
#1 SELECT4 rows · 3 ms

            

              
namestatustotal

              
                
Mira Okafordelivered128.40

                
Lucas Moreaushipped64.00

                
Hana Satopaid212.90

                
Tomás Ribeirodelivered39.99

              
            

          

          

            
JOINs from foreign keys

            
Type the start of a table name after JOIN and the relationship comes with it.

          

        


        

          

            

              
              
              
            

            
          

          

            
            

          

          

            
            
            
          

          

            
Type it back to break it

            
Every write needs a confirm click. DELETE without WHERE, DROP and
              TRUNCATE need the target typed back. Read-only connections are enforced by the database itself.

          

        

      

    


    
    

      

                
Most clients give you a tree and a grid.

        
Trees and grids are great for precise edits, and DBZuper has them too. It just doesn't stop there.

      

      

        

          You want to…
          The usual tree & grid
           DBZuper
        

        
Understand a schema you've never seenExpand 200 folders, one at a timeFly over it: clusters, tables, then columns

        
Find relationships nobody declaredHope someone added the foreign keysInferred from *_id naming, drawn dashed

        
Connect two distant tablesSketch it on paper, write the JOINs by handFind path to… and copy the JOIN

        
Chase one customer through the dataFive tabs, five WHERE clausesClick ports; share the trail as a URL

        
Write a JOINAutocomplete the table nameAutocomplete the whole ON clause

        
Avoid a 3 a.m. DELETE"Are you sure?" → EnterType the table back; read-only enforced by the server

        
Give your team accessInstall a desktop app on every laptopOne Docker image in a browser tab

      

    


    
    

      

                
And everything you'd expect.

      

      

        

          
          
Keyboard-first

          
One palette for tables, columns, records (users 42) and actions.

        

        

          
          
Server-level connections

          
Leave the database empty and browse every database on the server.

        

        

          

            MySQLMariaDBPostgreSQLSQLite
          

          
Four engines, one workspace

          
Timeouts, read-only and case-insensitive filters handled per engine.

        

        

          

            
            
            
            
          

          
Four themes × light and dark

          
Classic, Layered, Graphite and Radar. Dark by default, obviously.

        

        

          
          
History that keeps secrets

          
Per connection and global. Passwords in SQL are redacted before they're stored.

        

        

          
          
Yours, on your server

          
No external services at runtime. Credentials encrypted at rest. amd64 and arm64. MIT licensed.

        

      

    


    
    

      

        

                    
Some tables are bigger than others.

          
DBZuper is serious about your data and not very serious about itself. Hover or tap a huge table in the
            Universe and you'll see what we mean. Pin seven tables and something happens.

          

            
          

          
Pin them all. 0 / 7

        

        

          
          
order_items · row count

          
0

          
It's over 9,000!

        

      

    


    
    

      

        

                    
Running in one command.

          
    
            
  1. Run the image. One container, no external services.
    2. 
            
  2. Read the setup code from the log, so nobody else can claim your fresh install.
    3. 
            
  3. Click "Try it with demo data", or connect your own database.
    4. 
          

        

        

          

            
            

              
              
            

            
          


$ docker run -d --name dbzuper -p 8080:8080 \
    --add-host=host.docker.internal:host-gateway \
    -v dbzuper-storage:/app/storage \
    -v dbzuper-sqlite:/data/sqlite \
    ghcr.io/dbzuper/dbzuper:latest

$ docker logs dbzuper 2>&1 | grep "setup code"
  Setup code: ZUPR-7Q4K-M2XD

# open http://localhost:8080


        

      

    


    

      
      
One workspace. Every database.

      

        Get DBZuper 
        Star on GitHub
      

      
psst. type make a wish anywhere on this page, or .